Gilbert's chips, AFA (active feedback amplifier): AD830, LT6552, TDA856X

[jan.didden]

Originally Posted by AudioLapDance

This is true, however as abraxalito points out they may be linear enough (diminishing returns) and they shine where many conventional topologies start to stumble--high freq performance, in our cellular/wifi world.

Agreed, but one important factor here is 'we've done it always like this'. One design engineer once told me they HAD to call the '844 an 'opamp', otherwise nobody would buy it.
People buy opamps and not some other fancy-named contraption that they have no idea what it does, even if it would fit perfectly in their project.

jan

[scott wurcer]

Originally Posted by jan.didden

As to the AD830 and company, I believe that history shows us that Barrie's enthousiasm was not universially appreciated, to be diplomatic.
Maybe that is because it really is not THAT linear - you always need a certain voltage difference on the input pins and those input stages are not very linear.

jan

Jan you are quite correct, in practice the matching of the two halves of the AFA has limitations and these parts specified for reasonable yield rarely did better than .05 to .1% or 60-70 dBc. The AD844 is an op-amp BTW.

Current SOTA is -80dBc THD at 500MHz for amps at this time.

ALD - Please elaborate on what you mean by low level IMD in the absence of THD producing mechanisms.

[bcarso]

Originally Posted by scott wurcer

Current SOTA is -80dBc THD at 500MHz for amps at this time.

That's very impressive

[AudioLapDance]

Originally Posted by scott wurcer

Jan you are quite correct, in practice the matching of the two halves of the AFA has limitations and these parts specified for reasonable yield rarely did better than .05 to .1% or 60-70 dBc...

Can you say more?

Originally Posted by scott wurcer

...
ALD - Please elaborate on what you mean by low level IMD in the absence of THD producing mechanisms.

I was pulling quotes off of abraxalito's blog:
http://www.diyaudio.com/forums/blogs...d-project.html

abraxalito makes a high end chipamp from the TDA8561Q

Frank comments that "the THD figures are an order of magnitude worse than for the National chips, and the PSRR looks pretty dismal"

ab: "...the concept I am running hard here with is noise modulation. When I have a hypothesis I try my damndest to break it, flog it to death, test it in many ways to see how strong it is. This is one such test of the 'noise modulation is the missing SQ metric' hypothesis. So the high THD of this chip works in my favour..."

Frank: "I've used LTspice to look at Scott Wurcer's new baby, the Discrete Opamp, and Bob Cordell's "final", high performance power amp from his book. That's the order of performance I was getting, and what I'm looking for personally, as a safety margin..."

ab: "Order of performance for what? THD is the metric that Scott's using, he, like Doug Self pays little or no attention to what (IME) really matters for SQ which is low level IMD performance. So you're of the school that -100dB THD is a non-negotiable requirement for high-end sound? If so then there are a whole lot of high-end amps (considered subjectively) which violate that by huge margins! ..."



They go on to talk about PSRR, LM3886 and powersupplies, then:




ab:

"I agree that the power supply is the big issue, but only once the lower hanging fruit of the architecture (LTP or not LTP) has been plucked. That's low hanging because its easy to fix by just picking the right chip. Once the opamp is blameless then the next layer of that delicious yet eye-watering onion beckons... "

More can be found on his other write up:

Possibly the most frugal high-end sounding amp?

About chipamps he states:

ab: "...I have the hypothesis that what holds them back is the input stage, being as it is typically a LTP and a bipolar one at that. Added to that, they do tend to have inadequate PSRR by virtue of not having power supply separation between the signal and output stages. So they pollute their own patch, so to speak and filters can't be included because the necessary wires aren't brought off-chip...

...The search for a chipamp sans LTP has been running for a few weeks now and recently I've come upon some very interesting parts from Philips - the TDA856X series."

He then references Gilbert's two articles (Are Opamps Really Linear? and Nonlinear Effects of Radio-Frequency Interference in Opamps) and:

"The second paper is much more recent and covers real-world issues with LTPs in opamps which Barrie Gilbert doesn't address, maybe he wasn't even aware of. These issues have probably only come to light because of the greater degree of electromagnetic pollution in today's world - we live bathed in GSM and WiFi signals."

Nonlinear Effects of Radio-Frequency Interference in Opamps opampsusc_01.pdf

He then notes Charles Hansen's review of the PM48 (TDA8566, AFA chip)
http://www.marchandelec.com/ftp/revpm48.pdf

Talking about including the LTP in the feedback:

ab: "As to why they'd leave the LTP out of the loop - I think its perhaps because they haven't realized it is out of the loop. If they have realized then probably they think that its only an issue at DC - because offset is the only thing out of the loop - and haven't understood that RF is varying the offset voltage. Or that they think RF isn't an issue at audio. Just my speculation! "

He then goes on to do some power supply testing ...





abraxalito said he would stop in at this thread ...

[abraxalito]

Yep I'm still reading - interesting to hear myself quoted back again

Scott I'm not sure that your question is formulated in a manner I can respond to - of course the low level IMD won't be in the absence of THD producing mechanisms. However having linearity at high level (which is used for THD metric) isn't necessarily going to indicate excellent linearity at lower level (here I'm meaning below -30dB) because there can be a narrow dead band at zero without much effect on the full scale THD - an undithered quantizer at 16bits for example doesn't show up very well on a full-scale THD measurement.

[scott wurcer]

Originally Posted by abraxalito

Yep I'm still reading - interesting to hear myself quoted back again

Scott I'm not sure that your question is formulated in a manner I can respond to - of course the low level IMD won't be in the absence of THD producing mechanisms. However having linearity at high level (which is used for THD metric) isn't necessarily going to indicate excellent linearity at lower level (here I'm meaning below -30dB) because there can be a narrow dead band at zero without much effect on the full scale THD - an undithered quantizer at 16bits for example doesn't show up very well on a full-scale THD measurement.

A dead zone makes low level THD as well as IMD. By their nature classic differential input stages have no dead zone and the high level THD test is a decent metric to verify transfer linearity and slew performance. At least in my case the output stage for the specified load (600 Ohms) has a smooth transfer function around 0. The distortion performance is fine at -30dB.

Barrie Gilbert was concerned with factors that are inherent in the classic op-amp circuit. EMI, RFI, and power supply noise are decoupled in that they can yield to other better design techniques. In some cases circuits with virtually no PSRR can be made to work well. Maybe some folks are prepared to throw in the towel but I would propose that these issues can be dealt with outside of the choice of circuit architecture. That RF induced offset article injects RF directly into the input via a bias tee, hardly the situation in real applications. After all a lot of low level analog circuitry is working quite well (cap touch, audio, mems accelerometer) while your cell phone is blasting RF.

[abraxalito]

Originally Posted by scott wurcer

Maybe some folks are prepared to throw in the towel but I would propose that these issues can be dealt with outside of the choice of circuit architecture.

I have thrown in the towel as regards getting the SQ I'm looking for out of opamps and chipamps with LTP inputs. The TDA8566 is currently doing very nicely though so I'll be sticking with AFAs for the forseeable future.

That RF induced offset article injects RF directly into the input via a bias tee, hardly the situation in real applications. After all a lot of low level analog circuitry is working quite well (cap touch, audio, mems accelerometer) while your cell phone is blasting RF.

Is ADI going to follow in TI's footsteps and publish EMIRR reports on opamps? If so I'll be most curious to see the report on the AD797.

[scott wurcer]

Originally Posted by AudioLapDance

Can you say more?

For instance think about trying to track two things that have 10% open-loop linearity to better than say 1%, this would be .1% net mis-match. These circuits are pretty old BTW, the early instrumentation amplifiers relied on open-loop current transfer which is related to AFB. They also suffered from this problem. Check Note 2 below...

http://www.analog.com/static/importe...eets/AD521.pdf

[abraxalito]

AD603 and AD605 are also parts based on AFAs, though these have front-end attenuators which provide a volume control function. I have been using AD605 for well over a year now as I/V amplifier after passive I/V because it conveniently includes this volume stage, giving 40dB or so range of DC voltage controlled volume. It is though susceptible to RF and can't be directly connected to the I/V resistor of a TDA1545 without passive filtering in between if SQ is a priority. With the right filtering though it sounds awesome. Since its power supply is limited to 5.5V its best used to create a balanced output - this also will tend to cancel its high-ish 2H distortion at higher levels. Running balanced also keeps the noise down, which also is a bit marginal single-ended for a 16bit DAC application. I prefer to use it with a 50R source impedance - above this its getting a little noisy for my tastes though 100R is still quite usable in balanced.

http://www.analog.com/static/importe...eets/AD605.pdf

[scott wurcer]

Originally Posted by abraxalito

AD603 and AD605 are also parts based on AFAs, though these have front-end attenuators which provide a volume control function. I have been using AD605 for well over a year now as I/V amplifier after passive I/V because it conveniently includes this volume stage, giving 40dB or so range of DC voltage controlled volume.

Bushnel used it in one of their golf range finders.